Lair’s Dice by Self-Play

Recently I played around with the Counterfactual Regret Minimization algorithm combined with neural networks. I managed to get a bot that plays recently well at Liar's Dice.

I also made a website so you can try your luck against it at https://dudo.ai

I wanted to write a blogpost describing the details of the algorithm as well as how you can measure the convergence to the optimal Nash Equilibrium strategy.

title: s/Lair/Liar/